Twitter icon
Facebook icon
Google icon
Del.icio.us icon
LinkedIn icon
Andy Langton's picture

Co-occurrence, not occurrences: rethinking web copy for SEO

Are you writing your SEO copy for robots or people?

Keyword density is a phrase that (rightly) sends shudders down the spines of SEOs everywhere - even those who used to use keyword density as a basic way to assess their on-page targeting. So what's wrong with keyword density, and what should you be using instead?

What does keyword density even mean?

Keyword density is a very simple calculation - divide the number of keyword occurrences by the number of total words in a body of text. This tells you the percentage of times the keyword occurs. You'll hear numbers bandied around like 4% or even 40% as the 'ideal' keyword density.

To highlight the weaknesses of this approach, we only have to look at what most keyword density calculations miss:

  • Word roots and canonicals. If someone searches for "holidays", the canonical form is 'holiday'. If you don't account for this when checking density, your numbers will be wildly inaccurate
  • Synonyms. In most contexts, there's little difference (other than perhaps geography) in a search for 'vacation' vs. a search for 'holiday'
  • Hyponyms and hypernyms - these describe a more specific and less specific variant respectively. So, leisure is a hypernym for holiday while Christmas Eve is a hyponym
  • Google has it's own word relationship engine (use the ~ operator) that goes way beyond even these basic factors, because it accounts for both co-occurrence (which words are frequently placed together in text) and search history (which words do people search for when using a particular keyword). Google know that 'holiday' is frequently related to travel and flights - but also that Christmas is a holiday.

A keyword density tool that accounted for all of the above might perhaps be of interest. Unfortunately, the numbers it would spit out would be too vague to be of practical use - and how do you identify the right words anyway?

Scanning every book in the library

Remember the Google Books Library project? It's literally Google's attempt to read every book in the library, and store a digital copy of it. This gives them a body of text to analyse that has an in-built quality control. Books in the library have already undergone 'human review'. Analysing this text allows Google to understand how professional writing is formulated - and professional writers deliberately avoid word repetition. It also creates a body of "ham" text to compare to potential "spam" and train their algorithm - and this process will happen to your own copy.

Don't treat Google like it's stupid

How many web pages have you seen that use the same keyword, in the same form, in each of these areas?

  • Title
  • Meta Description
  • Meta keywords
  • URL
  • Heading 1, 2, 3 and beyond
  • In the first paragraph
  • Every x number of sentences

It's perfectly natural that the title of your page contains your keywords in order, and good URLs are likely to reflect your title. Hey, you can even put it in the H1 if your document has a very narrow focus. But give Google some credit - it knows your target keywords by that point. A barrage of keywords in all the 'right' places is not going to help you. In fact, it will create the impression that you used something like Web Position Gold to automate your on-page SEO.

Google is going to evaluate your copy in as intelligent a way as it possibly can - and that's going to involve comparing your copy to its own idea of professional/relevant copy. If you treat Google like its stupid, don't be surprised when it thinks your content is low quality.

Using this knowledge to create more relevant web copy

The more advanced Google becomes and the more text it is able to evaluate, the less useful outdated measures like keyword density and simple occurrences becomes. For some, this might seem like it makes a web copywriter's life more difficult. It should actually be the reverse. You can tell your copywriter to breathe a sigh of relief and go back to writing copy for people again. You don't need to constrain them to keyword x times in x paragraphs.You do need to ensure you account for word relationships and co-occurrence however. And of course, you need to know the keywords your audience use if you're trying to attract rankings.

Shameless plug: keyword discovery tool

We recently launched our own keyword discovery tool that we think can help you. Instead of relying solely on showing search volume for directly-related keywords, it shows words that co-occur, both in existing search results and in phrases that people search for.

So, a writer creating a holiday article can see immediately that words like holiday, cheap, deals, book, travel, inclusive, package, hotels, luxury, flights, breaks are all closely related to holidays, and may be suitable for use in text. We put these suggestions into word clouds, because we're ultra-modern and have fancy technology. Let us know any features you'd like to see!

Add new comment